AutoML Leaderboard

Best model	name	model_type	metric_type	metric_value	train_time	single_prediction_time
	1_Default_LightGBM	LightGBM	logloss	0.508137	13.23	0.0796
	2_Default_Xgboost	Xgboost	logloss	0.490055	12.62	0.0729
	3_Default_RandomForest	Random Forest	logloss	0.491796	17.63	0.1119
	8_LightGBM	LightGBM	logloss	0.48027	9.52	0.0658
	4_Xgboost	Xgboost	logloss	0.476665	11.62	0.065
	12_RandomForest	Random Forest	logloss	0.475697	23.34	0.145
	9_LightGBM	LightGBM	logloss	0.479837	9.39	0.0607
	5_Xgboost	Xgboost	logloss	0.495671	12.75	0.1229
	13_RandomForest	Random Forest	logloss	0.504447	20.24	0.1196
	10_LightGBM	LightGBM	logloss	0.492031	11.05	0.0648
	6_Xgboost	Xgboost	logloss	0.48139	13.18	0.07
	14_RandomForest	Random Forest	logloss	0.508036	28.63	0.1872
	11_LightGBM	LightGBM	logloss	0.510569	16.85	0.0673
	7_Xgboost	Xgboost	logloss	0.484476	14.62	0.0702
	15_RandomForest	Random Forest	logloss	0.492022	17.37	0.0913
	12_RandomForest_RandomFeature	Random Forest	logloss	0.479	21.57	0.1858
	16_RandomForest	Random Forest	logloss	0.478849	19.05	0.0918
	17_Xgboost	Xgboost	logloss	0.478176	13.6	0.0614
	18_Xgboost	Xgboost	logloss	0.478803	11.8	0.0721
	19_LightGBM	LightGBM	logloss	0.49038	11.3	0.065
	20_LightGBM	LightGBM	logloss	0.48027	11.44	0.063
	21_LightGBM	LightGBM	logloss	0.48027	14.62	0.0649
	22_Xgboost	Xgboost	logloss	0.480931	12.65	0.0681
	23_RandomForest	Random Forest	logloss	0.492364	18.68	0.1209
	24_Xgboost	Xgboost	logloss	0.47634	12.49	0.0641
	25_Xgboost	Xgboost	logloss	0.476992	13.54	0.0618
	26_Xgboost	Xgboost	logloss	0.476355	13.54	0.0739
	27_Xgboost	Xgboost	logloss	0.478084	13.44	0.0698
	28_LightGBM	LightGBM	logloss	0.479442	12.32	0.0662
	29_LightGBM	LightGBM	logloss	0.486012	11.5	0.0634
	30_LightGBM	LightGBM	logloss	0.478471	12.45	0.0661
	31_LightGBM	LightGBM	logloss	0.483768	16.85	0.0649
the best	Ensemble	Ensemble	logloss	0.470663	2.33	0.4231

AutoML Performance

AutoML Performance Boxplot

Features Importance

Spearman Correlation of Models

Summary of 10_LightGBM

<< Go back

LightGBM

• n_jobs: -1
• objective: binary
• num_leaves: 31
• learning_rate: 0.05
• feature_fraction: 0.9
• bagging_fraction: 0.9
• min_data_in_leaf: 20
• metric: binary_logloss
• custom_eval_metric_name: None
• explain_level: 1

Validation

• validation_type: kfold
• k_folds: 5
• shuffle: True
• stratify: True

Optimized metric

logloss

Training time

8.8 seconds

Metric details

	score	threshold
logloss	0.492031	nan
auc	0.839216	nan
f1	0.78492	0.338676
accuracy	0.764481	0.413397
precision	0.963855	0.930959
recall	1	0.0195094
mcc	0.535356	0.413397

Confusion matrix (at threshold=0.413397)

	Predicted as 0	Predicted as 1
Labeled as 0	629	286
Labeled as 1	145	770

Learning curves

Permutation-based Importance

Confusion Matrix

Normalized Confusion Matrix

ROC Curve

Kolmogorov-Smirnov Statistic

Precision-Recall Curve

Calibration Curve

Cumulative Gains Curve

Lift Curve

<< Go back

Summary of 11_LightGBM

<< Go back

LightGBM

• n_jobs: -1
• objective: binary
• num_leaves: 95
• learning_rate: 0.05
• feature_fraction: 0.9
• bagging_fraction: 0.8
• min_data_in_leaf: 10
• metric: binary_logloss
• custom_eval_metric_name: None
• explain_level: 1

Validation

• validation_type: kfold
• k_folds: 5
• shuffle: True
• stratify: True

Optimized metric

logloss

Training time

10.3 seconds

Metric details

	score	threshold
logloss	0.510569	nan
auc	0.827168	nan
f1	0.77322	0.296749
accuracy	0.750273	0.456609
precision	0.941748	0.922378
recall	1	0.0341478
mcc	0.503454	0.337968

Confusion matrix (at threshold=0.456609)

	Predicted as 0	Predicted as 1
Labeled as 0	647	268
Labeled as 1	189	726

Learning curves

Permutation-based Importance

Confusion Matrix

Normalized Confusion Matrix

ROC Curve

Kolmogorov-Smirnov Statistic

Precision-Recall Curve

Calibration Curve

Cumulative Gains Curve

Lift Curve

<< Go back

Summary of 12_RandomForest

<< Go back

Random Forest

• n_jobs: -1
• criterion: entropy
• max_features: 0.8
• min_samples_split: 50
• max_depth: 7
• eval_metric_name: logloss
• explain_level: 1

Validation

• validation_type: kfold
• k_folds: 5
• shuffle: True
• stratify: True

Optimized metric

logloss

Training time

21.3 seconds

Metric details

	score	threshold
logloss	0.475697	nan
auc	0.847216	nan
f1	0.785355	0.279841
accuracy	0.763934	0.443182
precision	0.982759	0.951497
recall	1	0
mcc	0.535712	0.443182

Confusion matrix (at threshold=0.443182)

	Predicted as 0	Predicted as 1
Labeled as 0	621	294
Labeled as 1	138	777

Learning curves

Permutation-based Importance

Confusion Matrix

Normalized Confusion Matrix

ROC Curve

Kolmogorov-Smirnov Statistic

Precision-Recall Curve

Calibration Curve

Cumulative Gains Curve

Lift Curve

<< Go back

Summary of 12_RandomForest_RandomFeature

<< Go back

Random Forest

• n_jobs: -1
• criterion: entropy
• max_features: 0.8
• min_samples_split: 50
• max_depth: 7
• eval_metric_name: logloss
• explain_level: 1

Validation

• validation_type: kfold
• k_folds: 5
• shuffle: True
• stratify: True

Optimized metric

logloss

Training time

18.1 seconds

Metric details

	score	threshold
logloss	0.479	nan
auc	0.846837	nan
f1	0.786384	0.420107
accuracy	0.767213	0.482636
precision	0.961538	0.936923
recall	1	0
mcc	0.539427	0.420107

Confusion matrix (at threshold=0.482636)

	Predicted as 0	Predicted as 1
Labeled as 0	648	267
Labeled as 1	159	756

Learning curves

Permutation-based Importance

Confusion Matrix

Normalized Confusion Matrix

ROC Curve

Kolmogorov-Smirnov Statistic

Precision-Recall Curve

Calibration Curve

Cumulative Gains Curve

Lift Curve

<< Go back

Summary of 13_RandomForest

<< Go back

Random Forest

• n_jobs: -1
• criterion: entropy
• max_features: 1.0
• min_samples_split: 20
• max_depth: 3
• eval_metric_name: logloss
• explain_level: 1

Validation

• validation_type: kfold
• k_folds: 5
• shuffle: True
• stratify: True

Optimized metric

logloss

Training time

17.7 seconds

Metric details

	score	threshold
logloss	0.504447	nan
auc	0.83027	nan
f1	0.780465	0.309195
accuracy	0.762295	0.548377
precision	0.941748	0.804861
recall	1	0.0298235
mcc	0.525067	0.548377

Confusion matrix (at threshold=0.548377)

	Predicted as 0	Predicted as 1
Labeled as 0	678	237
Labeled as 1	198	717

Learning curves

Permutation-based Importance

Confusion Matrix

Normalized Confusion Matrix

ROC Curve

Kolmogorov-Smirnov Statistic

Precision-Recall Curve

Calibration Curve

Cumulative Gains Curve

Lift Curve

<< Go back

Summary of 14_RandomForest

<< Go back

Random Forest

• n_jobs: -1
• criterion: gini
• max_features: 1.0
• min_samples_split: 30
• max_depth: 3
• eval_metric_name: logloss
• explain_level: 1

Validation

• validation_type: kfold
• k_folds: 5
• shuffle: True
• stratify: True

Optimized metric

logloss

Training time

26.5 seconds

Metric details

	score	threshold
logloss	0.508036	nan
auc	0.830146	nan
f1	0.787356	0.329314
accuracy	0.764481	0.483057
precision	0.907692	0.781043
recall	1	0.0569408
mcc	0.536524	0.329314

Confusion matrix (at threshold=0.483057)

	Predicted as 0	Predicted as 1
Labeled as 0	651	264
Labeled as 1	167	748

Learning curves

Permutation-based Importance

Confusion Matrix

Normalized Confusion Matrix

ROC Curve

Kolmogorov-Smirnov Statistic

Precision-Recall Curve

Calibration Curve

Cumulative Gains Curve

Lift Curve

<< Go back

Summary of 15_RandomForest

<< Go back

Random Forest

• n_jobs: -1
• criterion: gini
• max_features: 0.8
• min_samples_split: 50
• max_depth: 4
• eval_metric_name: logloss
• explain_level: 1

Validation

• validation_type: kfold
• k_folds: 5
• shuffle: True
• stratify: True

Optimized metric

logloss

Training time

15.1 seconds

Metric details

	score	threshold
logloss	0.492022	nan
auc	0.838868	nan
f1	0.786623	0.315955
accuracy	0.757377	0.58013
precision	0.928571	0.841134
recall	1	0.0294405
mcc	0.532983	0.315955

Confusion matrix (at threshold=0.58013)

	Predicted as 0	Predicted as 1
Labeled as 0	705	210
Labeled as 1	234	681

Learning curves

Permutation-based Importance

Confusion Matrix

Normalized Confusion Matrix

ROC Curve

Kolmogorov-Smirnov Statistic

Precision-Recall Curve

Calibration Curve

Cumulative Gains Curve

Lift Curve

<< Go back

Summary of 16_RandomForest

<< Go back

Random Forest

• n_jobs: -1
• criterion: gini
• max_features: 0.8
• min_samples_split: 50
• max_depth: 7
• eval_metric_name: logloss
• explain_level: 1

Validation

• validation_type: kfold
• k_folds: 5
• shuffle: True
• stratify: True

Optimized metric

logloss

Training time

16.9 seconds

Metric details

	score	threshold
logloss	0.478849	nan
auc	0.846858	nan
f1	0.786506	0.304496
accuracy	0.765574	0.473887
precision	0.959459	0.92248
recall	1	0.00263727
mcc	0.535418	0.38253

Confusion matrix (at threshold=0.473887)

	Predicted as 0	Predicted as 1
Labeled as 0	650	265
Labeled as 1	164	751

Learning curves

Permutation-based Importance

Confusion Matrix

Normalized Confusion Matrix

ROC Curve

Kolmogorov-Smirnov Statistic

Precision-Recall Curve

Calibration Curve

Cumulative Gains Curve

Lift Curve

<< Go back

Summary of 17_Xgboost

<< Go back

Extreme Gradient Boosting (Xgboost)

• n_jobs: -1
• objective: binary:logistic
• eta: 0.075
• max_depth: 7
• min_child_weight: 5
• subsample: 1.0
• colsample_bytree: 0.5
• eval_metric: logloss
• explain_level: 1

Validation

• validation_type: kfold
• k_folds: 5
• shuffle: True
• stratify: True

Optimized metric

logloss

Training time

11.2 seconds

Metric details

	score	threshold
logloss	0.478176	nan
auc	0.848818	nan
f1	0.784762	0.329198
accuracy	0.765574	0.425425
precision	0.961538	0.927465
recall	1	0.00859754
mcc	0.539143	0.413195

Confusion matrix (at threshold=0.425425)

	Predicted as 0	Predicted as 1
Labeled as 0	631	284
Labeled as 1	145	770

Learning curves

Permutation-based Importance

Confusion Matrix

Normalized Confusion Matrix

ROC Curve

Kolmogorov-Smirnov Statistic

Precision-Recall Curve

Calibration Curve

Cumulative Gains Curve

Lift Curve

<< Go back

Summary of 18_Xgboost

<< Go back

Extreme Gradient Boosting (Xgboost)

• n_jobs: -1
• objective: binary:logistic
• eta: 0.15
• max_depth: 7
• min_child_weight: 5
• subsample: 1.0
• colsample_bytree: 0.5
• eval_metric: logloss
• explain_level: 1

Validation

• validation_type: kfold
• k_folds: 5
• shuffle: True
• stratify: True

Optimized metric

logloss

Training time

9.7 seconds

Metric details

	score	threshold
logloss	0.478803	nan
auc	0.848502	nan
f1	0.786325	0.413766
accuracy	0.773224	0.466087
precision	0.986301	0.922577
recall	1	0.0175759
mcc	0.548602	0.466087

Confusion matrix (at threshold=0.466087)

	Predicted as 0	Predicted as 1
Labeled as 0	667	248
Labeled as 1	167	748

Learning curves

Permutation-based Importance

Confusion Matrix

Normalized Confusion Matrix

ROC Curve

Kolmogorov-Smirnov Statistic

Precision-Recall Curve

Calibration Curve

Cumulative Gains Curve

Lift Curve

<< Go back

Summary of 19_LightGBM

<< Go back

LightGBM

• n_jobs: -1
• objective: binary
• num_leaves: 31
• learning_rate: 0.1
• feature_fraction: 0.8
• bagging_fraction: 0.5
• min_data_in_leaf: 5
• metric: binary_logloss
• custom_eval_metric_name: None
• explain_level: 1

Validation

• validation_type: kfold
• k_folds: 5
• shuffle: True
• stratify: True

Optimized metric

logloss

Training time

9.0 seconds

Metric details

	score	threshold
logloss	0.49038	nan
auc	0.840738	nan
f1	0.782653	0.406291
accuracy	0.764481	0.406291
precision	0.933333	0.898995
recall	1	0.0126107
mcc	0.536515	0.406291

Confusion matrix (at threshold=0.406291)

	Predicted as 0	Predicted as 1
Labeled as 0	623	292
Labeled as 1	139	776

Learning curves

Permutation-based Importance

Confusion Matrix

Normalized Confusion Matrix

ROC Curve

Kolmogorov-Smirnov Statistic

Precision-Recall Curve

Calibration Curve

Cumulative Gains Curve

Lift Curve

<< Go back

Summary of 1_Default_LightGBM

<< Go back

LightGBM

• n_jobs: -1
• objective: binary
• num_leaves: 63
• learning_rate: 0.05
• feature_fraction: 0.9
• bagging_fraction: 0.9
• min_data_in_leaf: 10
• metric: binary_logloss
• custom_eval_metric_name: None
• explain_level: 1

Validation

• validation_type: kfold
• k_folds: 5
• shuffle: True
• stratify: True

Optimized metric

logloss

Training time

10.6 seconds

Metric details

	score	threshold
logloss	0.508137	nan
auc	0.828492	nan
f1	0.773453	0.390617
accuracy	0.752459	0.402512
precision	1	0.917652
recall	1	0.050816
mcc	0.51319	0.390617

Confusion matrix (at threshold=0.402512)

	Predicted as 0	Predicted as 1
Labeled as 0	609	306
Labeled as 1	147	768

Learning curves

Permutation-based Importance

Confusion Matrix

Normalized Confusion Matrix

ROC Curve

Kolmogorov-Smirnov Statistic

Precision-Recall Curve

Calibration Curve

Cumulative Gains Curve

Lift Curve

<< Go back

Summary of 20_LightGBM

<< Go back

LightGBM

• n_jobs: -1
• objective: binary
• num_leaves: 63
• learning_rate: 0.1
• feature_fraction: 0.5
• bagging_fraction: 0.8
• min_data_in_leaf: 50
• metric: binary_logloss
• custom_eval_metric_name: None
• explain_level: 1

Validation

• validation_type: kfold
• k_folds: 5
• shuffle: True
• stratify: True

Optimized metric

logloss

Training time

9.3 seconds

Metric details

	score	threshold
logloss	0.48027	nan
auc	0.848445	nan
f1	0.783685	0.277823
accuracy	0.766667	0.515542
precision	0.924779	0.859529
recall	1	0.0171264
mcc	0.533349	0.515542

Confusion matrix (at threshold=0.515542)

	Predicted as 0	Predicted as 1
Labeled as 0	698	217
Labeled as 1	210	705

Learning curves

Permutation-based Importance

Confusion Matrix

Normalized Confusion Matrix

ROC Curve

Kolmogorov-Smirnov Statistic

Precision-Recall Curve

Calibration Curve

Cumulative Gains Curve

Lift Curve

<< Go back

Summary of 21_LightGBM

<< Go back

LightGBM

• n_jobs: -1
• objective: binary
• num_leaves: 127
• learning_rate: 0.1
• feature_fraction: 0.5
• bagging_fraction: 0.8
• min_data_in_leaf: 50
• metric: binary_logloss
• custom_eval_metric_name: None
• explain_level: 1

Validation

• validation_type: kfold
• k_folds: 5
• shuffle: True
• stratify: True

Optimized metric

logloss

Training time

9.6 seconds

Metric details

	score	threshold
logloss	0.48027	nan
auc	0.848445	nan
f1	0.783685	0.277823
accuracy	0.766667	0.515542
precision	0.924779	0.859529
recall	1	0.0171264
mcc	0.533349	0.515542

Confusion matrix (at threshold=0.515542)

	Predicted as 0	Predicted as 1
Labeled as 0	698	217
Labeled as 1	210	705

Learning curves

Permutation-based Importance

Confusion Matrix

Normalized Confusion Matrix

ROC Curve

Kolmogorov-Smirnov Statistic

Precision-Recall Curve

Calibration Curve

Cumulative Gains Curve

Lift Curve

<< Go back

Summary of 22_Xgboost

<< Go back

Extreme Gradient Boosting (Xgboost)

• n_jobs: -1
• objective: binary:logistic
• eta: 0.075
• max_depth: 9
• min_child_weight: 10
• subsample: 0.8
• colsample_bytree: 0.6
• eval_metric: logloss
• explain_level: 1

Validation

• validation_type: kfold
• k_folds: 5
• shuffle: True
• stratify: True

Optimized metric

logloss

Training time

10.4 seconds

Metric details

	score	threshold
logloss	0.480931	nan
auc	0.848652	nan
f1	0.785375	0.308231
accuracy	0.769399	0.434951
precision	0.928571	0.930298
recall	1	0.0146874
mcc	0.543029	0.434951

Confusion matrix (at threshold=0.434951)

	Predicted as 0	Predicted as 1
Labeled as 0	647	268
Labeled as 1	154	761

Learning curves

Permutation-based Importance

Confusion Matrix

Normalized Confusion Matrix

ROC Curve

Kolmogorov-Smirnov Statistic

Precision-Recall Curve

Calibration Curve

Cumulative Gains Curve

Lift Curve

<< Go back

Summary of 23_RandomForest

<< Go back

Random Forest

• n_jobs: -1
• criterion: entropy
• max_features: 0.9
• min_samples_split: 30
• max_depth: 4
• eval_metric_name: logloss
• explain_level: 1

Validation

• validation_type: kfold
• k_folds: 5
• shuffle: True
• stratify: True

Optimized metric

logloss

Training time

16.0 seconds

Metric details

	score	threshold
logloss	0.492364	nan
auc	0.839283	nan
f1	0.783948	0.305427
accuracy	0.760656	0.540885
precision	0.94382	0.84216
recall	1	0.00558518
mcc	0.528235	0.411889

Confusion matrix (at threshold=0.540885)

	Predicted as 0	Predicted as 1
Labeled as 0	666	249
Labeled as 1	189	726

Learning curves

Permutation-based Importance

Confusion Matrix

Normalized Confusion Matrix

ROC Curve

Kolmogorov-Smirnov Statistic

Precision-Recall Curve

Calibration Curve

Cumulative Gains Curve

Lift Curve

<< Go back

Summary of 24_Xgboost

<< Go back

Extreme Gradient Boosting (Xgboost)

• n_jobs: -1
• objective: binary:logistic
• eta: 0.1
• max_depth: 6
• min_child_weight: 5
• subsample: 1.0
• colsample_bytree: 0.5
• eval_metric: logloss
• explain_level: 1

Validation

• validation_type: kfold
• k_folds: 5
• shuffle: True
• stratify: True

Optimized metric

logloss

Training time

10.2 seconds

Metric details

	score	threshold
logloss	0.47634	nan
auc	0.850071	nan
f1	0.788719	0.330144
accuracy	0.76776	0.50247
precision	0.979381	0.917908
recall	1	0.0111859
mcc	0.539531	0.330144

Confusion matrix (at threshold=0.50247)

	Predicted as 0	Predicted as 1
Labeled as 0	681	234
Labeled as 1	191	724

Learning curves

Permutation-based Importance

Confusion Matrix

Normalized Confusion Matrix

ROC Curve

Kolmogorov-Smirnov Statistic

Precision-Recall Curve

Calibration Curve

Cumulative Gains Curve

Lift Curve

<< Go back

Summary of 25_Xgboost

<< Go back

Extreme Gradient Boosting (Xgboost)

• n_jobs: -1
• objective: binary:logistic
• eta: 0.1
• max_depth: 8
• min_child_weight: 5
• subsample: 1.0
• colsample_bytree: 0.5
• eval_metric: logloss
• explain_level: 1

Validation

• validation_type: kfold
• k_folds: 5
• shuffle: True
• stratify: True

Optimized metric

logloss

Training time

11.2 seconds

Metric details

	score	threshold
logloss	0.476992	nan
auc	0.849939	nan
f1	0.783992	0.339966
accuracy	0.767213	0.46869
precision	0.975904	0.931539
recall	1	0.00901746
mcc	0.536719	0.407165

Confusion matrix (at threshold=0.46869)

	Predicted as 0	Predicted as 1
Labeled as 0	660	255
Labeled as 1	171	744

Learning curves

Permutation-based Importance

Confusion Matrix

Normalized Confusion Matrix

ROC Curve

Kolmogorov-Smirnov Statistic

Precision-Recall Curve

Calibration Curve

Cumulative Gains Curve

Lift Curve

<< Go back

Summary of 26_Xgboost

<< Go back

Extreme Gradient Boosting (Xgboost)

• n_jobs: -1
• objective: binary:logistic
• eta: 0.075
• max_depth: 6
• min_child_weight: 5
• subsample: 1.0
• colsample_bytree: 0.5
• eval_metric: logloss
• explain_level: 1

Validation

• validation_type: kfold
• k_folds: 5
• shuffle: True
• stratify: True

Optimized metric

logloss

Training time

10.9 seconds

Metric details

	score	threshold
logloss	0.476355	nan
auc	0.849995	nan
f1	0.786885	0.344069
accuracy	0.771038	0.484926
precision	0.987805	0.929789
recall	1	0.0129605
mcc	0.544515	0.440133

Confusion matrix (at threshold=0.484926)

	Predicted as 0	Predicted as 1
Labeled as 0	679	236
Labeled as 1	183	732

Learning curves

Permutation-based Importance

Confusion Matrix

Normalized Confusion Matrix

ROC Curve

Kolmogorov-Smirnov Statistic

Precision-Recall Curve

Calibration Curve

Cumulative Gains Curve

Lift Curve

<< Go back

Summary of 27_Xgboost

<< Go back

Extreme Gradient Boosting (Xgboost)

• n_jobs: -1
• objective: binary:logistic
• eta: 0.075
• max_depth: 8
• min_child_weight: 5
• subsample: 1.0
• colsample_bytree: 0.5
• eval_metric: logloss
• explain_level: 1

Validation

• validation_type: kfold
• k_folds: 5
• shuffle: True
• stratify: True

Optimized metric

logloss

Training time

11.0 seconds

Metric details

	score	threshold
logloss	0.478084	nan
auc	0.849438	nan
f1	0.783976	0.423512
accuracy	0.768306	0.519434
precision	0.976471	0.914805
recall	1	0.0154429
mcc	0.540981	0.423512

Confusion matrix (at threshold=0.519434)

	Predicted as 0	Predicted as 1
Labeled as 0	706	209
Labeled as 1	215	700

Learning curves

Permutation-based Importance

Confusion Matrix

Normalized Confusion Matrix

ROC Curve

Kolmogorov-Smirnov Statistic

Precision-Recall Curve

Calibration Curve

Cumulative Gains Curve

Lift Curve

<< Go back

Summary of 28_LightGBM

<< Go back

LightGBM

• n_jobs: -1
• objective: binary
• num_leaves: 15
• learning_rate: 0.05
• feature_fraction: 0.8
• bagging_fraction: 0.5
• min_data_in_leaf: 5
• metric: binary_logloss
• custom_eval_metric_name: None
• explain_level: 1

Validation

• validation_type: kfold
• k_folds: 5
• shuffle: True
• stratify: True

Optimized metric

logloss

Training time

10.2 seconds

Metric details

	score	threshold
logloss	0.479442	nan
auc	0.848294	nan
f1	0.782084	0.411295
accuracy	0.771038	0.486135
precision	0.969697	0.937055
recall	1	0.0188615
mcc	0.543058	0.486135

Confusion matrix (at threshold=0.486135)

	Predicted as 0	Predicted as 1
Labeled as 0	678	237
Labeled as 1	182	733

Learning curves

Permutation-based Importance

Confusion Matrix

Normalized Confusion Matrix

ROC Curve

Kolmogorov-Smirnov Statistic

Precision-Recall Curve

Calibration Curve

Cumulative Gains Curve

Lift Curve

<< Go back

Summary of 29_LightGBM

<< Go back

LightGBM

• n_jobs: -1
• objective: binary
• num_leaves: 15
• learning_rate: 0.2
• feature_fraction: 0.8
• bagging_fraction: 0.5
• min_data_in_leaf: 5
• metric: binary_logloss
• custom_eval_metric_name: None
• explain_level: 1

Validation

• validation_type: kfold
• k_folds: 5
• shuffle: True
• stratify: True

Optimized metric

logloss

Training time

8.9 seconds

Metric details

	score	threshold
logloss	0.486012	nan
auc	0.843523	nan
f1	0.783681	0.321103
accuracy	0.75847	0.529917
precision	0.952381	0.898691
recall	1	0.0249075
mcc	0.52653	0.321103

Confusion matrix (at threshold=0.529917)

	Predicted as 0	Predicted as 1
Labeled as 0	681	234
Labeled as 1	208	707

Learning curves

Permutation-based Importance

Confusion Matrix

Normalized Confusion Matrix

ROC Curve

Kolmogorov-Smirnov Statistic

Precision-Recall Curve

Calibration Curve

Cumulative Gains Curve

Lift Curve

<< Go back

Summary of 2_Default_Xgboost

<< Go back

Extreme Gradient Boosting (Xgboost)

• n_jobs: -1
• objective: binary:logistic
• eta: 0.075
• max_depth: 6
• min_child_weight: 1
• subsample: 1.0
• colsample_bytree: 1.0
• eval_metric: logloss
• explain_level: 1

Validation

• validation_type: kfold
• k_folds: 5
• shuffle: True
• stratify: True

Optimized metric

logloss

Training time

10.3 seconds

Metric details

	score	threshold
logloss	0.490055	nan
auc	0.839308	nan
f1	0.782692	0.333355
accuracy	0.763934	0.464718
precision	0.985915	0.937064
recall	1	0.0124038
mcc	0.53084	0.44653

Confusion matrix (at threshold=0.464718)

	Predicted as 0	Predicted as 1
Labeled as 0	651	264
Labeled as 1	168	747

Learning curves

Permutation-based Importance

Confusion Matrix

Normalized Confusion Matrix

ROC Curve

Kolmogorov-Smirnov Statistic

Precision-Recall Curve

Calibration Curve

Cumulative Gains Curve

Lift Curve

<< Go back

Summary of 30_LightGBM

<< Go back

LightGBM

• n_jobs: -1
• objective: binary
• num_leaves: 63
• learning_rate: 0.05
• feature_fraction: 0.5
• bagging_fraction: 0.8
• min_data_in_leaf: 50
• metric: binary_logloss
• custom_eval_metric_name: None
• explain_level: 1

Validation

• validation_type: kfold
• k_folds: 5
• shuffle: True
• stratify: True

Optimized metric

logloss

Training time

10.0 seconds

Metric details

	score	threshold
logloss	0.478471	nan
auc	0.849761	nan
f1	0.785356	0.40366
accuracy	0.768852	0.512269
precision	0.968421	0.908414
recall	1	0.0149315
mcc	0.541001	0.40366

Confusion matrix (at threshold=0.512269)

	Predicted as 0	Predicted as 1
Labeled as 0	700	215
Labeled as 1	208	707

Learning curves

Permutation-based Importance

Confusion Matrix

Normalized Confusion Matrix

ROC Curve

Kolmogorov-Smirnov Statistic

Precision-Recall Curve

Calibration Curve

Cumulative Gains Curve

Lift Curve

<< Go back

Summary of 31_LightGBM

<< Go back

LightGBM

• n_jobs: -1
• objective: binary
• num_leaves: 63
• learning_rate: 0.2
• feature_fraction: 0.5
• bagging_fraction: 0.8
• min_data_in_leaf: 50
• metric: binary_logloss
• custom_eval_metric_name: None
• explain_level: 1

Validation

• validation_type: kfold
• k_folds: 5
• shuffle: True
• stratify: True

Optimized metric

logloss

Training time

14.5 seconds

Metric details

	score	threshold
logloss	0.483768	nan
auc	0.845882	nan
f1	0.780803	0.280214
accuracy	0.762295	0.467221
precision	0.937173	0.879795
recall	1	0.0101233
mcc	0.526268	0.467221

Confusion matrix (at threshold=0.467221)

	Predicted as 0	Predicted as 1
Labeled as 0	661	254
Labeled as 1	181	734

Learning curves

Permutation-based Importance

Confusion Matrix

Normalized Confusion Matrix

ROC Curve

Kolmogorov-Smirnov Statistic

Precision-Recall Curve

Calibration Curve

Cumulative Gains Curve

Lift Curve

<< Go back

Summary of 3_Default_RandomForest

<< Go back

Random Forest

• n_jobs: -1
• criterion: gini
• max_features: 0.9
• min_samples_split: 30
• max_depth: 4
• eval_metric_name: logloss
• explain_level: 1

Validation

• validation_type: kfold
• k_folds: 5
• shuffle: True
• stratify: True

Optimized metric

logloss

Training time

15.6 seconds

Metric details

	score	threshold
logloss	0.491796	nan
auc	0.835751	nan
f1	0.787203	0.368453
accuracy	0.772131	0.514233
precision	1	0.890051
recall	1	0.021028
mcc	0.545397	0.514233

Confusion matrix (at threshold=0.514233)

	Predicted as 0	Predicted as 1
Labeled as 0	677	238
Labeled as 1	179	736

Learning curves

Permutation-based Importance

Confusion Matrix

Normalized Confusion Matrix

ROC Curve

Kolmogorov-Smirnov Statistic

Precision-Recall Curve

Calibration Curve

Cumulative Gains Curve

Lift Curve

<< Go back

Summary of 4_Xgboost

<< Go back

Extreme Gradient Boosting (Xgboost)

• n_jobs: -1
• objective: binary:logistic
• eta: 0.1
• max_depth: 7
• min_child_weight: 5
• subsample: 1.0
• colsample_bytree: 0.5
• eval_metric: logloss
• explain_level: 1

Validation

• validation_type: kfold
• k_folds: 5
• shuffle: True
• stratify: True

Optimized metric

logloss

Training time

9.5 seconds

Metric details

	score	threshold
logloss	0.476665	nan
auc	0.85012	nan
f1	0.788927	0.380297
accuracy	0.767213	0.545948
precision	0.966102	0.907328
recall	1	0.00959727
mcc	0.545609	0.380297

Confusion matrix (at threshold=0.545948)

	Predicted as 0	Predicted as 1
Labeled as 0	717	198
Labeled as 1	228	687

Learning curves

Permutation-based Importance

Confusion Matrix

Normalized Confusion Matrix

ROC Curve

Kolmogorov-Smirnov Statistic

Precision-Recall Curve

Calibration Curve

Cumulative Gains Curve

Lift Curve

<< Go back

Summary of 5_Xgboost

<< Go back

Extreme Gradient Boosting (Xgboost)

• n_jobs: -1
• objective: binary:logistic
• eta: 0.15
• max_depth: 6
• min_child_weight: 25
• subsample: 0.5
• colsample_bytree: 0.5
• eval_metric: logloss
• explain_level: 1

Validation

• validation_type: kfold
• k_folds: 5
• shuffle: True
• stratify: True

Optimized metric

logloss

Training time

9.9 seconds

Metric details

	score	threshold
logloss	0.495671	nan
auc	0.837126	nan
f1	0.787192	0.381811
accuracy	0.763934	0.381811
precision	0.896552	0.873917
recall	1	0.0142408
mcc	0.540949	0.381811

Confusion matrix (at threshold=0.381811)

	Predicted as 0	Predicted as 1
Labeled as 0	599	316
Labeled as 1	116	799

Learning curves

Permutation-based Importance

Confusion Matrix

Normalized Confusion Matrix

ROC Curve

Kolmogorov-Smirnov Statistic

Precision-Recall Curve

Calibration Curve

Cumulative Gains Curve

Lift Curve

<< Go back

Summary of 6_Xgboost

<< Go back

Extreme Gradient Boosting (Xgboost)

• n_jobs: -1
• objective: binary:logistic
• eta: 0.05
• max_depth: 9
• min_child_weight: 10
• subsample: 0.8
• colsample_bytree: 0.6
• eval_metric: logloss
• explain_level: 1

Validation

• validation_type: kfold
• k_folds: 5
• shuffle: True
• stratify: True

Optimized metric

logloss

Training time

11.0 seconds

Metric details

	score	threshold
logloss	0.48139	nan
auc	0.847923	nan
f1	0.78487	0.312795
accuracy	0.769945	0.44993
precision	0.95122	0.923801
recall	1	0.0141713
mcc	0.543344	0.44993

Confusion matrix (at threshold=0.44993)

	Predicted as 0	Predicted as 1
Labeled as 0	653	262
Labeled as 1	159	756

Learning curves

Permutation-based Importance

Confusion Matrix

Normalized Confusion Matrix

ROC Curve

Kolmogorov-Smirnov Statistic

Precision-Recall Curve

Calibration Curve

Cumulative Gains Curve

Lift Curve

<< Go back

Summary of 7_Xgboost

<< Go back

Extreme Gradient Boosting (Xgboost)

• n_jobs: -1
• objective: binary:logistic
• eta: 0.1
• max_depth: 4
• min_child_weight: 10
• subsample: 0.6
• colsample_bytree: 0.6
• eval_metric: logloss
• explain_level: 1

Validation

• validation_type: kfold
• k_folds: 5
• shuffle: True
• stratify: True

Optimized metric

logloss

Training time

11.9 seconds

Metric details

	score	threshold
logloss	0.484476	nan
auc	0.845528	nan
f1	0.787387	0.340812
accuracy	0.771585	0.521127
precision	0.957447	0.920725
recall	1	0.0175886
mcc	0.543389	0.521127

Confusion matrix (at threshold=0.521127)

	Predicted as 0	Predicted as 1
Labeled as 0	693	222
Labeled as 1	196	719

Learning curves

Permutation-based Importance

Confusion Matrix

Normalized Confusion Matrix

ROC Curve

Kolmogorov-Smirnov Statistic

Precision-Recall Curve

Calibration Curve

Cumulative Gains Curve

Lift Curve

<< Go back

Summary of 8_LightGBM

<< Go back

LightGBM

• n_jobs: -1
• objective: binary
• num_leaves: 95
• learning_rate: 0.1
• feature_fraction: 0.5
• bagging_fraction: 0.8
• min_data_in_leaf: 50
• metric: binary_logloss
• custom_eval_metric_name: None
• explain_level: 1

Validation

• validation_type: kfold
• k_folds: 5
• shuffle: True
• stratify: True

Optimized metric

logloss

Training time

7.4 seconds

Metric details

	score	threshold
logloss	0.48027	nan
auc	0.848445	nan
f1	0.783685	0.277823
accuracy	0.766667	0.515542
precision	0.924779	0.859529
recall	1	0.0171264
mcc	0.533349	0.515542

Confusion matrix (at threshold=0.515542)

	Predicted as 0	Predicted as 1
Labeled as 0	698	217
Labeled as 1	210	705

Learning curves

Permutation-based Importance

Confusion Matrix

Normalized Confusion Matrix

ROC Curve

Kolmogorov-Smirnov Statistic

Precision-Recall Curve

Calibration Curve

Cumulative Gains Curve

Lift Curve

<< Go back

Summary of 9_LightGBM

<< Go back

LightGBM

• n_jobs: -1
• objective: binary
• num_leaves: 15
• learning_rate: 0.1
• feature_fraction: 0.8
• bagging_fraction: 0.5
• min_data_in_leaf: 5
• metric: binary_logloss
• custom_eval_metric_name: None
• explain_level: 1

Validation

• validation_type: kfold
• k_folds: 5
• shuffle: True
• stratify: True

Optimized metric

logloss

Training time

7.3 seconds

Metric details

	score	threshold
logloss	0.479837	nan
auc	0.848114	nan
f1	0.787788	0.388107
accuracy	0.772131	0.517942
precision	0.955357	0.909906
recall	1	0.00980899
mcc	0.546227	0.393758

Confusion matrix (at threshold=0.517942)

	Predicted as 0	Predicted as 1
Labeled as 0	697	218
Labeled as 1	199	716

Learning curves

Permutation-based Importance

Confusion Matrix

Normalized Confusion Matrix

ROC Curve

Kolmogorov-Smirnov Statistic

Precision-Recall Curve

Calibration Curve

Cumulative Gains Curve

Lift Curve

<< Go back

Summary of Ensemble

<< Go back

Ensemble structure

Model	Weight
12_RandomForest	1
16_RandomForest	1
25_Xgboost	1
26_Xgboost	1

Metric details

	score	threshold
logloss	0.470663	nan
auc	0.853134	nan
f1	0.78835	0.368494
accuracy	0.770492	0.491177
precision	0.987179	0.920721
recall	1	0.00724617
mcc	0.542074	0.491177

Confusion matrix (at threshold=0.491177)

	Predicted as 0	Predicted as 1
Labeled as 0	676	239
Labeled as 1	181	734

Learning curves

Confusion Matrix

Normalized Confusion Matrix

ROC Curve

Kolmogorov-Smirnov Statistic

Precision-Recall Curve

Calibration Curve

Cumulative Gains Curve

Lift Curve

<< Go back